I always forget how to squash commits.

It should be a standard part of my workflow, but every time I need to clean up my history and combine a few commits, I never remember the process and since it might break badly, I freeze.

I finally wrote down a simple rule (and a mnemonic) to stop guessing.

An example scenario

I was working on a feature and ended up with a messy log. I wanted to combine the “typo” and “linter” fixes into the main feature commit, but I had already started some new work on top.

Here is my git log:

* 9a1b2c (HEAD) WIP: experimenting with new API  <-- LEAVE THIS ALONE
* 8f3a9b fix: typo in documentation              <-- SQUASH THIS
* 2c4e1d fix: linter error                       <-- SQUASH THIS
* 9b1a3f feat: add user authentication           <-- INTO THIS
* 7d2e1f refactor: database connection           <-- PARENT

I wanted to merge the middle fixes into feat: add user authentication.

The Solution

My instinct clashes with the fact that you don’t have to use the hash of the commit you want to modify (9b1a3f), but the one before 🤷.

The rule is: Always pick the Parent.

From now on, I will use the mnemonic R.I.P. messy history (Rebase Interactive Parent), hoping to remember it forever 💪.

# I want to modify 9b1a3f, so I rebase its PARENT (9b1a3f^)
git rebase -i 9b1a3f^

The Editor

While git log shows commits from the newest to the olders, the interactive editor does the opposite 🫠. So I shall remember to squash UP.

pick   9b1a3f feat: add user authentication
squash 2c4e1d fix: linter error
squash 8f3a9b fix: typo in documentation
pick   9a1b2c WIP: experimenting with new API
  1. Pick the oldest commit (the feature).
  2. Squash the commits below it.
  3. Pick the newer work at the end (so it stays safe).

It works on fixup commits too!

It is a silly thing, but it saves me from looking up the syntax every single time.